The Lightning Striking Probability for Offshore Wind Turbine Blade with Salt Fog contamination Qingmin Li,1,a) Yufei Ma,1 Zixin Guo,1 Hanwen Ren,1 Guozheng Wang,2 Waqas Arif,1 Zhiyang Fang,3 Wah Hoon Siew4 1State Key Lab of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Beijing, 102206, China 2School of Electrical Engineering, Shandong University, Jinan, 250061, China 3Sinomatech Wind Power Blade Co., Ltd, No.66 Xi Xiao Kou Rd., 100192, Beijing, China 4Department of Electronic & Electrical Engineering, the University of Strathclyde, Glasgow, G1 1XQ, UK The blades of offshore wind turbine are prone to be adhered with salt fog after long-time exposure in the marine-atmosphere environment, and salt fog reduces the efficiency of lightning protection system. In order to study the influence of salt fog on lightning striking probability (LSP), the lightning discharge process model for wind turbine blade is adopted in this paper considering the accumulation mechanism of surface charges around salt fog area. The distribution of potential and electric field with the development of downward leader is calculated by COMSOL Multiphysics LiveLink for MATLAB. A quantitative characterization method is established to calculate the LSP base on the average electric field before return stroke and the LSP distribution of blade is shown in the form of graphic view. Simulation results indicate that the receptor and conductor area close to receptor the area are more likely to get struck by lightning, and the LSP increases under the influence of salt fog. The validity of the model is verified by experiments. Furthermore, the receptor can protect the blade from lightning strikes effectively when the lateral distance between rod electrode and receptor is short. The influence of salt fog on LSP is more obvious if salt fog is close to receptor or the scope of salt fog area increases. I. INTRODUCTION The first offshore wind turbine of the world was installed in 1990 in Nogersund, Sweden. In the following 20 years, Denmark, Sweden, the Netherlands and the UK have built a number of demonstration offshore wind power projects which were mainly funded by the Government and research institutions. After dramatic growth of onshore wind power from first round of concession in 2003, China has also initiated the first round of national-level offshore wind farm concession projects in 2010. The turbine capacity increases from 567 MW in 2003 to nearly 13,803MW in 2009 in China1. It is foreseeable that offshore wind power development will boom dramatically in the future all over the world. a) Electronic-mail: [email protected]. 1 However, the damage caused by lightning striking remains the biggest threat to wind turbines2. Statistics show that about 5.56 wind turbine blades are damaged by lightning striking for every 100 blades per annum on average. The lightning protection for offshore wind turbines meets more severe challenges. Offshore wind turbines are installed in open environment with high tower and salt fog could adhere to the surface of blade easily under marine-atmosphere environment. The above factors will affect the protection efficiency of lightning protection system (LPS) for offshore wind turbine blades. In order to improve the protection efficiency of current LPS, it is vital to discover the area with high probability to be struck by lightning on blade surface. Field observation indicated that approximate 90% of lightning strikes on blade surface within 5m from blade tip3 in failure cases. Some researchers obtained the details of lightning propagation process, such as lightning leader characteristics by lightning local systems (LLSs), optical observation techniques and high-speed photography. Zeng4 summarized the recent progress on lightning physics and lightning protection research, including lightning observation, lightning initiation, lightning downward leader propagation, lightning attachment and return stroke, etc. Lu5 studied the connecting behavior of the downward and upward leaders during the attachment process preceding the first return stroke by high-speed video images. However, the observation methods also encountered challenges due to long experimental period and lack of the observation stations for wind power farms. Some researchers carried out long air gap experiments to study the distribution of attachment points on blade installed with different kinds of receptors6-8. Long9 analyzed the protection area of receptors based on self-consistent leader inception and propagation model. However, current researches are lack of the quantitative characterization for lightning striking probability on blade surface. In the marine-atmosphere environment, wind turbine blades are prone to be adhered with salt fog. Yokoyama10 found that soluble and non-soluble contaminants such as dust, sand and salt deposited on the blade surface would enhance the surface discharge activity. The research carried by Douar11 indicated that pollution deposit on the surface of insulator reduces the flashover voltage, irrespective of the polarity of applied voltage. Kumar12 found that flashover voltage reduced with the increasing of salt deposit density on GFRP material, and the flashover voltage under negative switching impulse is less compared with positive switching impulse. However, the influence mechanism of salt fog in various positions and scopes to the distribution of lightning attachment points is not clear. This paper discusses the quantitative characterization of lightning striking probability (LSP) distribution of wind turbine blade and the influence of salt fog on the LSP based on the achievements from charged particles 2 accumulation and long gap discharge areas. It will be a theoretical basis on making effective lightning protection measures and painting advanced marine antifouling coatings for offshore wind turbine blades. II. LIGHTNING DISCHARGE PROGRESS MODEL FOR WIND TURBINE BLADE Les Renardierers Group proposed that the natural lightning discharges showed some similarities with the laboratory long gap discharges based on the experimental studies13-16 and the observation results obtained by K. Berger17, given as follows. 1) Both share the same physical process. The development of a natural lightning is initialized and sustained by the formation of ‘streamer corona’ and ‘leader’ discharges, very similar to those observed in the laboratory long gap discharges18; 2) The lightning attachment point is influenced by the development of both the downward and upward leaders; 3) For both natural lightning discharge and laboratory long gap discharge, the critical electric field for initializing an upward leader is roughly the same, with the approximate value of 500kV/m. Actually, it is impossible to conduct a natural lightning discharge experiments in the lab due to restricted space and lack of background conditions. Based on the similarity theory proposed by Les Renardierers Group, the laboratory long gap discharge experiments are widely used to study the attachment process for natural lightning discharges. Two typical test setups recommended in IEC 61400-2419 are appropriate for the tests on the complete blades used for design development and verification. Each test arrangement is intended to initialize electrical activities such as corona, streamers and leaders, at the targeted specimen just before a lightning attachment may happen. Yokoyama20 used a 12MV high voltage impulse generator at Shiobara testing yard of Central Research Institute of Electric Power Industry to investigate the lightning attachment manner to the wind turbine blades. Three-meter long blade-sample was cut from an actual twelve-meter long wind turbine blade made of GFRP. He found that the discharges progressed along the surface of a blade-sample tip and attached to the receptor in many cases, and sometimes the discharges passed through the air and attached to the receptor directly. Radičević21 used the typical 3MW 1:40-scaled wind turbine model, with an arching high-voltage electrode under different modes of stationary and rotating blades, to study the physical mechanism of the blade rotation influence on the wind turbine in triggering lightning discharges. It was found that, the breakdown voltage decreased and the connection point of the leader approached the blade tip with an increasing blade speed. Considering the above research advances, the laboratory long gap discharge experiment can be an efficient method to investigate the attachment process for the natural lightning discharges. 3 A. Simplified model of wind turbine blade The structure of wind turbine blade is shown in Fig. 1(a). The blade is made of glass fiber reinforced plastic (GFRP) and equipped with LPS including a tip receptor and copper stranded conductor. Generally, the length of operational offshore wind turbine blade is more than 45m, and the computation complexity will increase if a whole blade model is adopted. In order to reduce computation complexity and time, a simplified wind turbine blade is adopted as shown in fig. 1(b). The blade body of simplified model is made of GFRP, with relative permittivity of 4.0, and the size is 400×200×3mm. An aluminum-made receptor is installed at the tip of blade, with the size of 20×200×3mm, and it is connected to the ground through a copper conductor. Salt fog could be easily adhered to the blade surface, and the main constituent of the salt fog is water soluble salt, such as NaCl. Increasing salt fog contamination will attach to the wind turbine blade after long-time exposure in the marine-atmosphere environment, and the shape of the salt fog contamination renders random style. In order to facilitate the modeling process while ensuring universality and generality of the proposed model, a cylindrical salt fog contamination model located at the coordinates (0.1, 0.05) is adopted in the paper, as shown in Fig. 2. Setting the center point of the underside of blade as the starting point of coordinates, a coordinate system is established in this paper, as shown in Fig. 3. The height of external electrode of 0.2m, the distance between the electrode bottom and blade upper surface is assumed to be 0.4m in the simulation model.